This paper explores how entrepreneurs benefit from their networks to create and implement innovations and, more specifically, delves into the specific characteristics and patterns of their network usage. We depart from previous work on the contribution of networks to the development of innovations which is clearly biased towards multinational firms and high-tech start-up firms and instead focus on a broad sample of established small- and medium-sized firms. The networks of any entrepreneur can be described in terms of: i) their diversity, ii) the strength of the various relationships, iii) their directedness and iv) the availability of structural holes. Our empirical contribution contains a survey shedding a light on how the network ties of the entrepreneur/small business owner contribute to the development of innovations. Our cross-industry sample included 594 entrepreneurs and small business owners in the Netherlands who developed at least one innovation in the past three years. By telephone they were asked about their network usage of all kinds of other parties to support the innovation process. For their most recent innovation questions were asked into the nature, roles and network characteristics of the involved network partners. According to the data collected, small firms appear to be very different in their use of networks to support the innovation process. As hypothesized, we find that radical innovations require heavy network use in terms of diversity, strength of relationships, directedness and structural holes. Hierarchical cluster analysis reveals six patterns of network use to support the innovation process, varying from the supplier-based pattern, characterised by a low use of network partners and involving innovations which are initiated and delivered by suppliers without major modifications, to the system-based pattern with a kind of network use mostly found with large-scaled, radical types of innovations with a major impact on many firms. Although we can draw some conclusions about particular network benefits, many issues about the most ‘optimal’ network contribution to innovation remain unsolved. Contingencies can play an important role: for instance the purpose and the social context have a large impact on the particular role network partners can play. The larger the uncertainty and search for the exact nature of an innovation the more important new information is. A diverse network with many weak ties, directed ties and structural holes is expected to play a more beneficial role in such circumstances.